Mass media advertising: Information or wallpaper? - 6. Noticing specific ads

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Mass media advertising: Information or wallpaper?

Smit, E.G.

Publication date

1999

Link to publication

Citation for published version (APA):

Smit, E. G. (1999). Mass media advertising: Information or wallpaper?. Het Spinhuis.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s)

and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open

content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please

let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material

inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter

to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You

will be contacted as soon as possible.

Noticing specific ads

The central question in this chapter is: 'To what extent do people notice specific advertisements, and to what extent is this related to characteristics of the advertisements (such as product category), and to user characteristics (such as beliefs, attitudes and behaviour towards advertising)?' The results of three different studies will be used to answer this question. In Section 6.1, part of the results of the nation-wide survey (study II) is presented to illustrate how little respondents were able to recall a television commercial during the telephone interview. Different measures are compared by means of an experiment (study III) which is described in Section 6.2. The aim of this experiment is to test whether 'recognition' would be a better method than 'recall' for measuring the amount of magazine advertisements and television commercials respondents claimed seeing before. In Section 6.3, results of the fourth face-to-face study on noticing magazine advertisements are used to analyse which advertisements are more recognised than other advertisements. Finally, LISREL analysis is used in Section 6.4 to test the relations between the recognition of specific advertisements, the respondents' claimed advertising behaviour and other user characteristics, such as beliefs, attitudes, medium use and demographics.

6.1 Retrieval problems (study II)

At the end of the telephone interview of study II, all respondents (n=1065) were asked what they recall of the (at that time) new advertising campaign for the Dutch beer Grolsch. They were first asked for what beer brand they recently saw an advertising campaign ('brand recall'1)- Almost 40% of the respondents spontaneously recalled seeing

the campaign for Grolsch2. The others named other brands (33.8%),

described an advertisement without naming a brand (5.8%), or stated that they were not aware of any recently advertised beer brand (20.6%). The respondents who did not spontaneously recall the brand (n=641), received an extra question and were asked whether they recalled recently seeing an advertisement for Grolsch ('aided brand recall'3). More than

40% of these respondents said 'yes', which resulted in a 'total brand recall' of 65.1% of the sample (n=693) (see Table 6.1).

Table 6.1: Brand recall Grolsch (n=1065)

n

_%

unaided brand recall 424 39.8

aided brand recall 269 25.3

total brand recall 693 65.1

These 693 respondents were then asked in what medium (or media) they saw this campaign4. The respondents who recalled seeing the new

commercial on television (n=567) were then asked to recall the theme (pay-off) of the campaign or describe the commercial ('ad recall'5). The

same question was asked to the respondents who recalled seeing the advertisement in a newspaper (n=86). Only 28 respondents accurately remembered the pay-off of the television commercial ( 'Komt tijd, komt Grolsch')6, 34 respondents named the 'old' pay-off which is also referred

to in this campaign ('Vakmanschap is meesterschap'), and 70 respondents only referred to content elements of the commercial. The others (n=211) described another commercial for the same brand. Only 7 respondents remembered the text of the newspaper advertisement, 6 respondents named some content elements and 18 respondents described another advertisement for Grolsch (see Table 6.2).

Table 6.2: Ad recall Grolsch (% of sample, n=1065)

television commercial

(a) recalled seeing the new ad on television/in a newspaper (b) recall pay-off 1 ('Komt tijd, komt Grolsch')

(c) recall pay-off 2 ('Vakmanschap is meesterschap') (d) recall one or more content elements 7

total recall of specific ad ('proven' ad recall) [b+c+d]

n _% 567 53.2 28 2.6 34 3.2 70 6.6 132 12.4 newspaper ad n

_%

86 8.1 2 0.2 5 0.5 6 0.6 13 1.2

Figure 6.1 pictures the results of brand recall and ad recall. The 'funnel' effect is striking: only 12.4% of the respondents (n=1065) actually recalled elements of the new Grolsch television commercial and 1.2% recalled elements of the newspaper advertisement.

sample n=1065

T

Dran

unaided brand recall

yes: n=424 *• " ^ no: n=641

I

aided brand recall

^V

yes: n = 269 no: n=372

t

total brand recall

I

medium? television

t

ad recall

^

newspaper ad recall

no: n=224 yes: n=343 yes: n=31 no: n=55

'proven' ad recall: n=13 1.2% of the sample 'proven' ad recall: n=132

12.4% of the sample

Figure 6.1: Funnelling from 100% to less than 15% of the sample

In other words: it appeared difficult for the respondents to recall the television commercial of the new advertising campaign. While the majority of respondents recalled recently seeing an advertisement for Grolsch when given the product category as a cue, only 12.4% was able to 'prove' this recall by naming the pay-off or describing other elements of the television commercial. A comparison of the recall scores of the target group 'men and beer users' (n=375) with the scores of the rest of the sample, showed no significant differences. Recalling the newspaper advertisement seemed even more difficult.

The question is whether these figures are also found in other studies. Since 1984, all television commercials in South Africa have been tracked by means of a telephone survey of 200 respondents a week ( DuPlessis, 1994a). For each television commercial, ad recall measures are taken two or three weeks after its first appearance. This has resulted in a comprehensive database of some 10,000 commercials. DuPlessis found that the norm (average recall) for a commercial after three weeks on the

air is about 14%. Based on these South Africa data and the 8% norm in England, DuPlessis8 assumes that the previous result of 12.4% for The

Netherlands is not exceptional. However, an ad recall percentage of 12.4% does reflect the difficulty of those respondents who at first recalled seeing the new advertising campaign and were later not able to describe any elements of the commercial. The question is how this effect can be explained. One possible explanation is the media schedule. In Figure 6.2, this schedule is presented by means of the grey bars.

48 49 50 Week (1995)

Recall (lines): Media input (bars):

f1'"1 -total ad recall tv commercial • 4 5 " television commercial < » ~ proven ad recall tv commercial | 1 / 1 newspaper and magazine ad

(3) — total ad recall print ads

<*> — proven ad recall print ads

Figure 6.2: Media schedule and recall figures per week

The Grolsch campaign started December 2, 1995 (week 48) with a newspaper ad (1/1 page) and some short teasers on television. Broadcasting of the 4 5 " television commercial started in week 49. The data from the telephone survey were conducted over nine weeks. The percentages shown (the four lines) are based on the number of respondents who recalled seeing the new Grolsch ad ('total ad recall' or 'proven ad recall') divided by the number of respondents in that week9.

In week 52 no telephone calls were made because of the holiday period. The figures shown for this week are the average scores of weeks 51 and 1. Figure 6.2 shows that the total ad recall curve for television does not follow the media input: after week 5 1 , the curve declines in stead of increases. The curve of proven ad recall on the other hand, is more a result of media scheduling: during the first burst, ad recall increases until week 2, and decreases thereafter until the next burst in week 5. This difference between the curves can be caused by the scheduling of another Grolsch commercial during the same period. During the interview period it was clear that some respondents recalled the

commercial for Grolsch winter beer, which was on the air in December 1995.

More importantly, it is possible that a lot of respondents were actually exposed to the specific commercial, but were not able to recall it. There are several causes for this retrieval problem. In the first place, the first study (see Chapter 4) showed that advertising is relatively unimportant for most people and thereby not well elaborated 10. Also the

confusion due to two campaigns at the same time makes this elaboration even more difficult (Kent, 1994). Secondly, retrieval of information is inhibited when the wrong cue is given during the interview. In this study, the interviewer asked the respondent to describe the commercial ('ad recall'). Because most advertising enters the mind via the eyes, a recall measure tries to establish traces of the memories that were formed mainly visually (DuPlessis, 1994a). In other words, the potential visual storage of the Grolsch commercial had to be retrieved from memory by means of a verbal cue given by the interviewer at the other end of the telephone line11. Visual prompting - as used in

recognition measures - could have fitted the memory traces better. Both recall and recognition measures involve the ability of an advertisement to have attracted attention and to have entered memory. After comparing both methods, DuPlessis (1994a, p.79, 1998) concludes that the difference between the two methods lies in the cueing material to prompt memory: recognition shows the commercial (or a version of it), while recall is a verbal prompt. The recognition method was introduced by Daniel Starch in the twenties. Under Starch, the technique of 'noted scores' became a syndicated service in 1933. In it, respondents were asked whether they 'noted' advertisements in a magazine and, if so, whether they 'read most' of it (Dunn, 1990). Shortly thereafter, George Gallup developed a measure of print recall. Gallup did not show people the ads themselves. Instead he asked them a series of questions to ascertain whether they had or had not seen an advertisement (Biel, 1996). The discussion about which of these measures is best, has been going on for forty years (see for instance Gibson, 1983; Krugman, 1977). Recognition is said to overestimate the number of advertisements seen ('overclaim'), especially for disliked material. Recall on the other hand is criticised for being overly based on words (arguments) instead of emotions, images, or music (Hansen, 1995; Hoogerbrugge, 1997).

6.2 Recall vs. recognition (study III)

In a subsequent study, we further elaborate on the retrieval problem by conducting an experiment in which the difference between recall and recognition of television commercials and magazine advertisements was tested. This experiment was partly based on Zielske's (1982) studies on the recollection of different types of advertisements (see Chapter 3).

The aim of our experiment (study III) is to compare different measures for noticing advertisements in two different media. Two groups of 40 respondents were asked whether they noted the magazine advertisement or television commercial for a specific brand12. The first group (rh=40)

was first given the brand as a cue and then asked to describe the ad ('proven recall'). The same ad was shown to the second group (n 2=40) in

a masked form. This latter group was asked whether they had seen the ad before, and if so, to fill in the missing brand name ('proven recognition'). Each respondent answered the questions for eight magazine advertisements and eight television commercials. The results of these two different measures are summarised in Table 6.3.

Table 6.3: Recall versus recognition (% of the respondents with correct cl aim)

recall (ni= =40) recognition (n2 =40)

magazine television magazine television

0 of 8 ads 87.5% 27.5% 7.5% 2.5% 1 of 8 ads 10.0% 32.5% 12.5% 7.5% 2 of 8 ads 2.5% 20.0% 17.5% 22.5% 3 of 8 ads 0.0% 15.0% 22.5% 20.0% 4 of 8 ads 0.0% 5.0% 15.0% 32.5% 5 of 8 ads 0.0% 0.0% 12.5% 12.5% 6 of 8 ads 0.0% 0.0% 10.0% 2.5% 7 of 8 ads 0.0% 0.0% 2.5% 0.0% 8 of 8 ads 0.0% 0.0% 0.0% 0.0% > 0 ads 12.5% 72.5% 92.5% 97.5% > 4 ads 0.0% 0.0% 25.0% 15.0%

This table shows that for the magazine advertisements, 87.5% of the respondents recalled none of the eight advertisements, compared with 7.5% that were not able to recognise one of the ads and fill in the missing brand. A difference of 80 percent points or 32 respondents. A smaller difference (of 25 percent points) occurred for remembering the television commercials: 27.5% recalled none of the commercials compared with 2.5% that did not correctly name the missing brand. In other words, measuring noticing ads by means of recognition results in more correct claims. This difference between the recall and recognition figures is especially apparent for magazine advertisements. When measured by means of recognition, 92.5% of the respondents had at least one of the magazine advertisements correct ('> 0 ads'). When measured by recall, only 12.5% had at least one correct claim. For television commercials, this difference is much smaller (97.5% and 72.5%, respectively). This media difference also emerged for a more strict criterion of more than half of the advertisements correct ('> 4 ads').

To test the previous observed differences o'n an average level, a Paired Samples T-test was conducted. This test shows that the average number of noted television commercials (mean=1.43) differs

significantly from the number of noted magazine advertisements (mean=0.15) when measured by means of recall ( t(39)=-6.20, p=0.00). This is not the case when measured by means of recognition (mean=3.15 and 3.20 respectively, t(39)=-2.21, p=0.84). To test whether recognition results in higher correct claims than recall, the average scores of the two measurements are compared. An Independent Samples T-test shows that both groups differ significantly in noticing both types of advertising. Moreover, the difference in average scores is higher for magazine advertisements (difference: 3.0; t(78)=-9.49, p=0.00) than for television commercials (difference: 1.77; t(78)=-10.08, p=0.00).

In this study, some other variables were measured which were expected to influence the respondents' recognition or recall of advertisements. These variables are the respondents' general awareness of a brand in a given product category ('brand awareness'), the likeability of the advertisements and their use of the advertised brand ('brand use'). An analysis of covariance showed that the main effect of measurement type on noticing magazine advertisements is still significant when controlled for these three variables ( F(l,75)=30.61, p=0.00). Of these variables, only the interaction effect of 'brand awareness' is significant ( F ( l , 75)=17.81, p=0.00). The more respondents are aware of the measured brand, the more they recognised the magazine advertisements. For noticing television commercials, the main effect of type of measurement is significant as well (F(l,75)=18.82, p=0.00) when controlled for 'brand awareness', 'likeability' and 'brand use'. Only the interaction effect of likeability is significant (F(l,75)=21.54, p=0.00), indicating that the more respondents like the commercials, the more they recognised them.

In conclusion, a measurement of the number of specific advertisements noted by means of recognition resulted in significantly higher correct claims than a measurement by recall, especially for magazine advertisements. This effect of measurement type is still significant when 'brand awareness' (for magazine advertisements) and ' likeability' (for television commercials) are taken into account. This means that the difference between recognition and recall is especially apparent for magazine advertisements. Awareness of the brand name enhances this effect, possibly because familiarity with the brand name draws the reader's attention to the ad while she is skimming the magazine. For television commercials the difference between recognition and recall is smaller - but significant -, and enhanced by a positive evaluation of the specific commercial. As was mentioned in the previous chapter, an evaluation in terms of liking and disliking is especially appropriate for using a medium with external pacing. While the commercials are being displayed to the viewer, it will be looked at when liked. If the commercial is evaluated as bad, it will be avoided.

6.3 Noticing magazine advertisements (study IV)

Based on the previous results, recognition can be considered as an appropriate method for measuring whether respondents noted specific magazine advertisements. Validation of the copy-test (which is based on recognition) by means of camera observation shows that this method is fairly accurate (Bronner, 1987; Brown, 1990; Bunn, 1982; Cebuco,

1985; Kent, 1994; Walter, 1995).

As a result, 'noticing magazine advertisements' was measured in our next and final study as having seen the advertisement before ('recognition') and being able to prove this recognition by filling in the missing part of the advertisement ('proven recognition'). This final study (study IV) was conducted in a face-to-face situation in which the copy-test was used (see Chapter 3). Following the procedure of this method, all respondents were asked to go through their most read magazine title, and to indicate for each advertisement whether they had seen that ad before ('recognition')13. Results show that all respondents were shown at

least two and at the most seventeen advertisements (dependent on the number of advertisements in the selected magazine). A total of 671 advertisements were shown to 88 respondents, or about eight advertisements per respondent. Of all advertisements shown, on average 4 4 . 1 % were recognised (or 296 ads). This is about 5 percent points lower than the average recognition scores for women's magazine advertisements, as measured in the standardised questionnaire of RRO 14

(Admedia, 1997), and about 10 5 percent points lower than the average recognition score of 53% for television commercials (DuPlessis, 1994a, p.86). Furthermore, the respondents were asked to fill in the missing brand name ('proven recognition'). Of all ads recognised, 25.3% (or 75 ads) were 'proven' recognised by filling in the correct brand name. This is 11.2% of all advertisements shown. In Table 6.5, the percentage of ads that are recognised and proven recognised are presented per product category (for more details about the advertisements, see Chapter 3).

Table 6.5: (Proven) recognition of advertisements per product category

recognition (re) proven recognition (pr)

product category (pc): (% yes) (% correct brand name)

body care products (n=105 ads) 54.3% 19.0%

medical products (n=36 ads) 50.0% 8.3%

fast moving consumer goods (n=172 ads 45.3% 7.6%

durable consumer goods (n=302 ads) 39.7% 10.9%

financial services (n=51 ads) 37.3% 11.8%

all product categories (n=666 ads) 43.8% 11.3%

test results: re by pc x!(4)=8.33, p=0.08; pr by pc X!(4)= =9.09, p=0.06

This table shows that advertisements for body care products are most recognised as well as most proven recognised. Ads for medical products and fast moving consumer goods are also more recognised than average. Remarkably, with respect to 'proven recognition', brands for financial services - such as banks and insurance companies - are relatively more recalled. Brands for fast moving consumer goods, for instance, were more difficult to recall than brands for the other product categories. The relationships between product category and both dependent variables are, however, not significant.

To explain why some advertisements are more noted than others , we compared advertisements for products which are used with those which are not (variable 'product use'). We also made this comparison for advertisements which are of particular interest to the respondent (variable 'product-ad interest'), and advertisements that are liked (variable 'likeability') (Table 6.6).

Table 6.6: (Proven) recognition by product use, product interest and likeability

la. non product users (n=251 ads) lb. product users (n=420 ads)

2a. not product interested group (n=430 ads) 2b. product interested group (n=241 ads) 3a. negative group (likeability < 5.5; n=194 ads) 3b. positive group (likeability > 5.5; n=460 ads) all groups (n=671 ads)

recognition (re) proven recognition (pr)

(% yes) (% correct brand name)

45.0% 8.0% 43.6% 13.1% 40.9% 9 . 1 % 49.8% 14.9% 24.7% 3.6% 53.0% 14.8% 44.1% 11.2%

test results of re by la,b: x'(l)=0.13, p=0.72; pr by la,b: x!(l)=4.2, p=0.04; re by 2a,b: x!(D=0.4.92, p=0.03; pr by

2a,b: x2(l)=5.36, p=0.02; re by 3a,b: x'(l)=44.22, p=0.00; pr by 3a,b: x2W=16.78, p=0.00.

In this table, the results of three comparisons are presented. The first comparison shows that the group users (group l b ) only differ significantly from the group that does not use the advertised product or service (group la) when asked to fill in the missing brand name ('proven recognition', right column). Both groups recognise the same amount of advertisements. For the other comparisons, it shows that the ones who are interested in the advertised product (group 2b) and the ones who are positive about the advertisement (group 3b; likeability > 5.5) recognise significantly more of the advertisements shown. Moreover, they could fill in more missing brand names than the others. ' Likeability' was originally measured as a score ( ' 1 ' is very negative, '10' is very positive). On average, the advertisements (n=654) were given 6 out of 10, which is a B minus in the United States. Those who recognised the ads, gave on average 6.6, while the others gave significantly one point less (t(652)=-7.78, p=0.00). For 'proven recognition' a comparable difference occurred (6.8 versus 5.9; t(652)=^.45, p=0.00). A logistic regression analysis is used to see which of the variables could predict a (proven)

recognition score best of all. This analysis showed that only on the basis of the 'likeability' scores, a prediction of 'recognition' and 'proven recognition' could be made (see Table 6.7).

Table 6.7: Results of two logistic regression analyses (partial r) (n=654 ads) dependent variables:

predictors: recognition proven recognition

product use (d) -0.03 0.00

product interest (d) 0.05 0.03

likeability 0 . 2 3 * 0.18»

model statistics: X'(3)=64.23, p=0.00 X!(3)=35.15, p=0.00

correct class fication: 65.0% correct classification: 88.5% d = dummy variable (1 = present 0 = absent), * = signi icant at the 0.05 level.

To sum up

This study showed that on average 4 4 . 1 % of all advertisements shown were recognised by the respondents. Approximately a quarter of these recognised ads were also 'proven recognised' by filling in the correct brand name. Advertisements that are liked were more noted than other ads. It should be noted, however, that the direction of the relation between likeability and (proven) recognition is not necessarily as described. Since the aim of this study is to explain whether advertisements are noted or not, likeability is used as a predictor. The direction of causality could also be the reverse; that is, familiarity with an ad may be attributed to greater liking for it (as suggested by the mere exposure hypothesis of Zajonc, 1970).

6.4 Explaining noticing specific ads (study IV)

In the previous section, the amount and type of noted advertisements were described and explained by some advertisement variables. The central question in this section is to what extent recognising specific advertisements can be explained by user characteristics. These user characteristics were discussed in the previous chapter: advertising use (measured by three statements about what the respondent would do when confronted with magazine advertising, and the question whether he or she generally pays attention to magazine advertisements, see Table A8.3 in the appendix), evaluation of advertising in terms of beliefs (measured by nine statements about functions of magazine advertising, see Table A8.4 in appendix) and attitude (whether someone is very positive, positive, neutral, negative or very negative about magazine advertising), medium use (the number of titles the respondent reads regularly, whether he or she has a subscription to one or more titles, and the type of magazines he or she likes to read: 'news', 'women' or 'other') and some demographics (sex, age, education, household size and income). In the appendix, the average scores as well as the results

of the factor analyses are shown. Results show that the respondents are moderately positive about magazine advertisements in terms of their general use (they 'sometimes' or even 'regularly' look at them or pay some attention to them) and beliefs, especially about the information function of magazine advertising. Moreover, respondents evaluate magazine advertising more or less as 'positive' (their average score on the attitude variable is 3.42 on a five-point scale).

To test whether the previous user characteristics are directly or indirectly related to 'recognition' of specific ads15, the analysis

procedure of Chapter 5 was used. First, the different statements were reduced to one or more factors (see Table A8.3 and Table A8.5 in the appendix). Second, all dependent variables were individually related to the other variables by means of backward regression analyses (see Table A8.6). Third, the model based on the (not excluded) relationships is tested by means of a LISREL analysis. The fitting model is presented in F i g u r e 6 . 3 ( x2( 4 5 ) = 5 8 . 3 7 , p = 0 . 0 9 , A F G I = 0 . 8 1 ; R2 r ec o g nl ti o n = 0 . 1 3 ,

R2u s e ~ 0 ' 3 4 , R2| i k j n g = 0 . 1 9 , R2entertainment—0.13, R information=0• 2 6 , R2irrjtation=0.34). The explained variance of 'recognition' is rather low,

while 'information', 'irritation' and 'advertising use' are explained relatively well by this model's variables.

household note: values are standardised path coefficients; _{only significant relations are shown (p<0.05)}

recognition

household size

L regular r e a d i n g

Figure 6.3: E x p l a i n i n g r e c o g n i t i o n of magazine a d v e r t i s e m e n t s ( n = 8 8 )

This figure shows four major relationships. First, recognition of specific advertisements is positively related to the respondents general behaviour towards magazine advertisements ('use') and to their magazine readership ('regular reading'). In other words: the more respondents claim that they pay attention to magazine advertisements and the more titles these respondents read regularly, the more

advertisements they recognised. This means that noticing specific magazine ads ('recognition') is predicted by other medium/advertising use and only indirectly by the other variables.

Second, advertising use is (directly) influenced by the respondents' beliefs about the irritation value of magazine advertising and the attitude towards magazine advertising (degree of liking): the more positively respondents evaluate magazine advertising, the more they claim to pay attention to magazine advertisements. 'Advertising use' is also positively influenced by the number of titles respondents read.

Liking is the third important variable that is explained in this model. This variable appeared - as predicted - to be significantly explained by the belief factors 'information', 'entertainment' and 'irritation'. The more respondents evaluate magazine advertising as informative, entertaining, and nor irritating, the more positive they are about magazine advertising in general. Besides these beliefs, 'liking' is also significantly related to household income: the higher the income, the more positive respondents are about magazine advertising. The direction of the relation between income and liking is contrary to the results of the meta-analysis (see Figure 2.3 in Chapter 2). This difference could be caused by the focus on television in these studies instead of magazine readership.

Fourth, this model shows that primarily age, household characteristics (income and size) and the medium use variable 'regular reading', are important in explaining the three belief factors. These general user characteristics indirectly influence (general and specific) advertising use via opinions about advertising in a particular medium.

To sum up

It was generally shown that respondents who were positive about magazine advertising (in terms of their claimed advertising use), recognised more advertisements than others. Moreover, those who regularly read magazines recognised more ads as well. Although these relationships were significant, they could only explain 13% of the variance in recognition. Attention to magazine advertising (claimed advertising use) appeared better explained (34%), namely by the model's variables: the respondents general evaluation of magazine advertising (liking of magazine advertising, beliefs about the entertaining and information function of magazine advertising and the absence of irritation) and other user characteristics (age and household characteristics).

A comparison of this model (Figure 6.3) with the exploratory model for magazine advertising of the survey (study II, see Figure 5.4) shows a few differences. First, this alternative model of the fourth study showed the more important role of the demographics age and household characteristics. Second, the beliefs about information and entertainment of magazine advertising are indirectly related to advertising use via liking. This indirect effect was at first expected and could not be

accepted in the second study. Since the alternative model (Figure 6.3) is based on a rather small and not randomly drawn sample (n=88), rejection of the exploratory model of study II (Figure 5.4) is a bit premature. As well as sample differences, a few other differences exit between the two studies. The formulation of some of the belief statements are slightly changed in the fourth study (see Chapter 3) and the questionnaire was completed in the presence of the interviewer in stead of by telephone or mail (mode differences).

6.5 Conclusion

The question to what extent respondents remembered their exposure to advertisements was answered in three steps. First, the results of the telephone survey (study II) were used to illustrate the difficulty respondents had in searching their memories for traces of a specific advertising campaign. While the majority of respondents said that they had recently seen the new advertising campaign for a Dutch brand of beer, only 12% were able to recall the pay-off or content elements of the television commercial. One of the possible causes for this retrieval problem is the data gathering method (i.e. telephone interviewing). By means of this method, the visual storage of the television commercial had to be retrieved from memory by means of a verbal cue instead of visual prompting.

Second, results of the third study were presented to examine the difference between verbal cueing ('recall') and using visual cues ('recognition'). In this experiment, the recollection of specific television and magazine advertisements was measured by means of the two measures 'recall' and 'recognition'. The claims of the respondents were checked by asking them to fill in the missing brand name. The results showed that 'recognition' resulted in more correct claims than 'recall'. The difference between the two measures was especially visible in measuring the recollection of magazine advertisements.

Finally, the results of the fourth face-to-face study were used to explain why some magazine advertisements were more noted than other advertisements, and why some people note more magazine advertisements than others. It was shown that likeable advertisements were more noted (or: noted and thus evaluated as more likeable). Analyses also showed that respondents who regularly read magazines, recognised more advertisements than others. Moreover, respondents who are positive about magazine advertising in terms of their advertising use recognised more specific advertisements. In other words: respondents who claim to read or pay attention to magazine advertisements in general, do this apparently more than others. This general behaviour (i.e. advertising use) could to some extent (namely 34%) be explained by these respondents' general evaluation of magazine advertising (beliefs and attitude) and other user

characteristics, such as age and household characteristics (income and size).

The previous chapter showed that part of the variance in advertising use could be explained by evaluation of advertising in that medium and some of the other user characteristics. At the end of Chapter 5 the question was posed to what extent this claimed advertising behaviour could be interpreted as an indicator of (actual) behaviour in stead of one of the components of the attitude construct. This question refers to the research question 'To what extent is (claimed) advertising use related to awareness of specific advertisements?'. Besides describing different measures of ad memory - (proven) recall or (proven) recognition -, this sixth chapter presented the results of the fourth study in which advertising use was related to the recollection of having seen the advertisements before. This study showed that the recognition of magazine advertisements was only moderately but significantly -related to advertising use (r=0.30, p<0.05). When other variables were taken into account it was shown that their relationship was 0.23 (standardised path coefficient based on the LISREL analysis). In other words: the recollection of specific advertisements appeared only moderately related to people's (claimed) general behaviour. Possible explanations of this effect will be discussed in the next chapter.

Notes chapter 6

1 Generating the brand name when given the product category as cue is called 'brand recall' (Bendapudi, Linville, Mishra & Singh, 1994).

2 The respondents were able to name a maximum of three brands. Of all the brands spontaneously named, Grolsch was second. The top five of 'Top Of Mind Awareness' (TOMA) is: Heineken (50.7%), Grolsch (39.8%), Amstel (23.2%), Bavaria (18.3%), and Dommelsch (13.3%).

3 Asking whether a respondent saw a campaign in a given period or recently when giving the brand name as a cue is called 'aided brand recall' (Pieters & Bijmolt,

1994).

4 This was a multiple response question. Number of responses per medium: 'television' (567), 'newspaper' (86), 'radio' (30), 'magazine' (66), 'outdoor' (37), 'free cards' (1), 'direct (non) mail' (10), 'cinema' (30), 'other' (17).

5 'Ad recall' is the ability to remember (parts of) an ad after giving the brand as a cue. Different names are given to this concept, dependent on the specificity of cues given or answers required: 'brand prompted recall' (DuPlessis, 1994a), 'proven recall' (Dunn & Ziff, 1974), 'related recall' (Dunn & Ziff, 1974; Hodock, 1980; Stewart & Furse, 1986) or 'main idea recall' (Riter, Balducci & McCollum, 1982).

6 Old Grolsch campaigns referred to old values such as 'workmanship' and 'old trades'. The recent campaigns refer to a new Zeitgeist: there is a time to ... (referring to choices in life), and (thus) a time to drink Grolsch. The old pay-off

'Vakmanschap is meesterschap' means 'craftsmanship is mastery'. The new

pay-off 'Komt tijd, komt Grolsch' means 'When the time comes: Grolsch'.

7 Content elements of the television commercial are: father and son (scenes: at the hairdresser, police station, giving first condom, boy sleeping in car, boy in tree, hotel room), little boy grows older, meeting after years (hotel room) and the tune (by The Byrds: 'Turn, turn, turn'). Content elements of the newspaper ad are: different small pictures, words about time ('Time to ...'), green letters in pay-off/theme/sentences.

8 Based on contact by e-mail (November 27, 1997), see also: DuPlessis (1994a). 9 Number of respondents per week: n=118 (week 49), n=121 (week 50), n=122

(week 51), n=113 (week 1), n=131 (week 2), n = 112 (week 3), n = 110 (week 4), n=142 (week 5), n=96 (week 6).

10 This elaboration process is described by the Elaboration Likelihood Model of Petty and Cacioppo (1986). The hypothesis is that the more motivated a person is, the more centrally the information is processed and the more enduring the information effect will be.

11 The remark that recall by means of the telephone survey was rather difficult for the respondent is illustrated by the influence of the interviewers. As explained in Chapter 3, the training of the interviewers was part of Pondman's study (see Section 3.3). In general, differences in most variables of the survey could not be attributed to differences in interviewer training. However, 'unaided brand recall' and recall of the television commercial ('total ad recall') differ significantly in the three interviewer conditions. Respondents who were interviewed by motivated

interviewers were more inclined to say 'yes' than the respondents who were interviewed by interviewers from the control group ('unaided brand recall': 43.6% in the motivation condition compared to 41.6% in the task condition and 34.4% in the control condition, x2(2)=6.48, p=0.04; 'ad recall': relatively more respondents recalled the commercial when interviewed by interviewers of the task condition (64.6%) or motivation condition (63.5%), compared with the control group (52.4%), x2(2)=5.5, p=0.04). The interviewers with special training were possibly more motivated to encourage the respondents to search their'memories for information about the campaign.

12 No demographic differences were found for the two measurement groups. 13 Not all advertisements in the magazine titles were used (see Chapter 3 for

selection of advertisements).

14 'RRO' means 'Reclame Reactie Onderzoek' ('Advertising Reaction Research', translated by author). RRO is a standardised instrument developed byAdmedia. This instrument measures the advertising strength of magazine advertisements in terms of reach ('recognition'), reactions (by means of 'thought listing') and likeability statements based on the Viewer Response Profileof Schlinger (1979). The percentages listed are based on the magazine titles Libelle' and 'Margriet'. 15 The user characteristics were measured for each respondent. 'Noticing specific

advertisements' was measured by means of 'recognition' as well as 'proven recognition' of several advertisements (see Section 6.3). Because the analysis is now at the respondent instead of the advertisement level, an average 'recognition score' had to be computed per respondent. This score was calculated by dividing the number of correct claims (meaning the number of 'yes' answers) by the number of advertisements shown, which resulted in a variable with scores between 0 and 1 (mean=0.44, skewness=0.21, kurtosis=-0.50). The more strict criterion of 'proven recognition' is not used in this section. Results showed that almost half of the respondents were not able to fill in one of the missing brand names. Moreover, this variable is too skewed to use as dependent variable in linear regression analyses (skewness=2.27, kurtosis=6.86). In order to correct for this extreme skewness, 'proven recognition' was coded into five classes (skewness=0.61, kurtosis=-0.75). Regression analysis (method: Backward) showed that this variable was only predicted by 'regular reading' (5=0.26, p<0.05; R2=0.06, F(l,77)=5.55, p=0.02).